Visa warns AI is making scams harder to spot and easier to scale

Artificial intelligence is making criminals more productive, too. Visa says fraudsters are using it to write convincing phishing messages, find security weaknesses, and expand their operations into more countries without needing to hire more people to do the work.

The company’s Fall 2026 Biannual Threats Report examines activity observed from January through June. It describes criminals using AI to improve familiar attacks, giving them faster ways to research targets, refine malicious code, and test payment defenses.

For consumers, that means a scam might arrive in fluent English, reference a real purchase, and send you to a website that looks convincing. The badly written email was already an unreliable way to identify fraud, and AI is making that approach even less useful.

“AI is changing the economics of fraud by giving criminals the ability to move faster, operate across borders and adapt their tactics in real time,” said Paul Fabara, Visa’s chief risk and client services officer. He said the company is combining technology, network intelligence, and human expertise to identify emerging threats earlier.

One of the biggest concerns involves the companies behind the businesses people actually recognize. Payment processors, eCommerce platforms, and hospitality technology providers can serve many customers, making a successful attack against one provider especially valuable to criminals.

Visa recorded a 62 percent increase in incidents involving third-party agents compared with the previous six months. Those providers were linked to approximately 56 percent of the compromised payment account numbers identified during the reporting period.

The company also reported an increase of more than 1,200 percent in payment account numbers recovered from data breaches. That measures the credentials Visa recovered, rather than an equivalent increase in money stolen or the number of breaches.

Hospitality providers were a recurring target, with criminals attacking reservation systems used by multiple travel and accommodation businesses. In one investigation, attackers deployed payment-stealing code in both a provider’s hosted software and customer environments running its software locally.

For anyone booking a vacation, another finding may be more unsettling. Visa describes fraudulent messages that impersonate booking platforms and reference actual reservation details to make the request appear legitimate.

Those messages direct travelers to fake websites seeking additional payments for supposed room taxes, city taxes, reservation confirmation, or special requests. A message that knows where you are staying is much easier to believe than a random demand for money, particularly when you are already expecting updates about your trip.

This is where the combination of stolen information and convincing presentation becomes dangerous. Recognizing your hotel and travel dates in a message can make you lower your guard, even though the sender could have obtained those details through a compromise.

Visa Scam Disruption identified more than $1.4 billion in potential fraud during the six-month period, up 40 percent from the preceding period. Its investigations covered investment scams, impersonation, deceptive online commerce, unauthorized recurring billing, and fraudulent merchant networks operating across multiple regions.

That $1.4 billion represents potential or attempted fraud, not confirmed consumer losses. The report also does not establish that AI was involved in every incident included in its figures.

Criminals are also taking advantage of attention around sporting tournaments, concerts, and other major events. Visa describes operations using AI-generated multilingual content, advertising, social media, and convincing fake websites to reach potential victims across different markets.

There is some encouraging news in the report, although the comparison periods require careful reading. Payment account enumeration, which involves automatically guessing and testing card details, increased 35 percent compared with the same period in 2025 but declined 7.8 percent compared with the previous six months.

Visa’s Risk Operations Center blocked 200 percent more enumeration transactions than in the preceding six-month period. The report separately cites a 21.4 percent annual decline in losses associated with enumerated credentials for Visa’s fiscal first quarter of 2026, which covered October through December 2025.

These figures show that more attempted attacks do not automatically translate into more successful theft. They also make it difficult to reduce the findings to a claim that payment security is getting worse across the board.

Looking ahead, Visa warns that more capable AI agents could help criminals automate complex sequences of attacks and continuously test fraud controls. Those systems could adjust their tactics as defenses respond, reducing the amount of human supervision needed to keep an operation running.

Visa has a commercial interest in promoting its own security technology, and its report reflects the activity it observes and investigates. It should be read as a view into that activity, rather than a complete accounting of worldwide cybercrime.

Still, the reservation scams illustrate the problem without needing any dramatic predictions. Criminals can use accurate personal details to earn someone’s trust, and AI can help them package that deception for more people, in more languages, with less effort.

☕

Support independent tech journalism

NERDS.xyz is independently owned and operated. If you enjoy my coverage of Linux, AI, hardware, cybersecurity, and tech culture, consider supporting the site on Ko-fi.

Support NERDS.xyz
Written by

Brian Fagioli ✔

Technology journalist and founder of NERDS.xyz

Brian Fagioli is a technology journalist and founder of NERDS.xyz. A former BetaNews writer, he has spent over a decade covering Linux, hardware, software, cybersecurity, and AI with a no nonsense approach for real nerds.

Leave a Comment